Analysis for Effect of Mutation Subtypes on the Allele Frequency Spectrum and Population Genetics Inference

Step 1) Download zipped analysis directory from: https://zenodo.org/record/7443570#.Y5t2yS1h1pQ. Use tar -xvzf afs_analysis.tar.gz to unzip file. Major files in directories include:

a) ./data

• ./data/mst_sfs/: Text file for each of 96 subtypes. File all_subtypes_afs.txt contains all 96 AFS in one file
• ./gw_sites_info.txt: Text file containing each variant info genome wide used in analysis
• ./all_MST_100kb_counts.txt: Text file containing data for 100Kb window analysis (D, mst counts, etc)
• ./data/DaDi/bootstrap_SFS: 100 bootstrapped AFS for each subtype
• ./data/simulated_neutral/all_*.txt: Simulated neutral AFS using theta estimates for A_C.AAA and C_T.ACG

b) ./output

• sigma_mat.txt: Large sigma matrix used for covariance calculations in D-2 statistic
• ./output/D2_stat/by_subtype: Output of D-2 statistic for each subtypes AFS
• ./output/DaDi: Output from running dadi 10 times on each subtypes AFS.

c) ./scripts

• entire_analysis.R: R script containing entire pipeline sourcing each relevant analysis file
• afs_paper_functions.R: R script with functions used
• Note: files with accompanying same name .slurm scripts were run in parallel to avoid long computation times.

Step 2) Install software needed to run all analysis completely from scratch.

• R libraries needed can be found in ./scripts/libraries_needed.R.
  • Note: Analysis run using R version 4.2.2)
  • Note: Package 'xlsx' was used in analysis to export to .xlsx files, but some computers had issues installing so commented out
• Python 3
• DaDi - Python 3 version (https://dadi.readthedocs.io/en/latest/user-guide/installation/)
• SLURM for parallel computing

Step 3) Run analysis by going through entire_analysis.R in sections.

• Create AFS for each of 96 mutation subtypes. Outputs separate txt file for each and a single with all 96 subtypes.
• Create dataframe of genome wide AFS statistics for each of 96 subtypes. Includes D, D-2, % singles, doubles, triples, etc AFS heterogeneity across subtypes and signals of recurrent mutations and gene conversion shaping the AFS
  • Signals of mutation rate/recurrent mutations on AFS. Look at singleton/doubleton ratio by mutation rate
  • Signals of gBGC on AFS. Split into WS, SW, and indifferent mutation types. Compare D and D-2 across subtypes
• DaDi Analysis: Effect of genome wide AFS heterogeneity across subtypes on demographic inference
  • These scripts require installing DaDi on Python3 (see above).
• Regional 100Kb Analysis: Effect of subtype composition and local genomic factors on the regional AFS